1 / 26

Arjan Durresi, Raj Jain, Gojko Babic, Justin Dolske

97-1089: Modifications to Appendix B and Sections 3.1.7 and 3.2.7 of Testing Baseline Text on Scalable Configurations. Arjan Durresi, Raj Jain, Gojko Babic, Justin Dolske The Ohio State University Contact: Jain@cse.ohio-state.edu http://www.cse.ohio-state.edu/~jain/. Overview.

babu
Download Presentation

Arjan Durresi, Raj Jain, Gojko Babic, Justin Dolske

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. 97-1089: Modifications to Appendix B and Sections 3.1.7 and 3.2.7 of Testing Baseline Text on Scalable Configurations Arjan Durresi, Raj Jain, Gojko Babic, Justin Dolske The Ohio State UniversityContact: Jain@cse.ohio-state.edu http://www.cse.ohio-state.edu/~jain/

  2. Overview • Why these modifications? • Methodology • Examples

  3. Scalable Configurations • ATM testing equipment are expensive. • Scalable Configurations permit to simulate the desired basic configuration using a limited number of generators. • But there are many ways to set up the scalable connections configurations and the results could vary with the set up.

  4. P1 In P3 In P2 In P4 In P3 Out P2 Out P4 Out P1 Out Module 1 Module 2 A Sample 4-to-4 Configuration • Different implementations could provide different results. • P1-P2-P3-P4-P1 Four module crossings • P1-P3-P4-P2-P1 Two module crossings

  5. In Out In Out Module 1 Module 2 In Out In Out Problem w/ Current Text • Some switches set only bidirectional VCsCan’t have the same VCI on the same port for two VCs. In Out In Out Module 1 Module 2 In Out In Out

  6. VCC Chain • Performance testing requires setting up connections between ports of the switch. • Some connections are internal through the switch fabric and others are external through wires or fibers. • An external connection between two switch ports is referred in this appendix as a wire W. • The sequence of concatenated connections (internal and external) is called a VCC Chain. • The proposed algorithm permits to create standard VCC Chains for any number of generators and any number of ports Þ Scalable and basic (both)

  7. P4 P2 P3 P1 P1 P4 P2 P3 W1 W2 W3 Example • 4-to-4 configuration with one generator IN OUT Module 1 Module 2 One of six possible VCC chains: P1 W1 W2 W3 P1

  8. VCC Chain Implementation • Implementation of External Connections 1. Numbering the ports 2. Identifying the ports connected to generators and analyzers 3. Numbering the wires • Implementation of Internal Connections

  9. P23 P15 P16 P24 P25 P17 P26 P18 P19 P27 P28 P20 1. Port Numbering Module # • Need to group modules by technology and speed • This port numbering helps creating VCC chains that cross modules using a simple algorithm P11 P13 1) P1 P3 P5 P7 P9 P14 Group 1 2) P2 P4 P6 P8 P10 P12 3) Group 2 P21 P22 4) Group 3

  10. 2. Generator/Analyzer Ports • Identifying the ports connected to the generator and/or analyzers • Avoid having only one port left over in a group. (That port cannot be connected externally to any other port) • This condition does not apply if the switch allows loopbacks. • Note: The algorithm works with loopbacks also.

  11. 3. Numbering the Wires • In each group start with the first output port available. Connect it to the next port whose input is available. (Note: With loopbacks, the output of a port can be connected to the input of the same port. The rest of the methodology is same.) • Continue until all ports have been connected. • Numbered the wires sequentially as W1, W2, …with the restriction that the end of wire Wi and the beginning of W(i+1) must be different ports. • May need to skip some wires and include them in the next round.

  12. IN OUT Background Traffic Generator 2 Foreground Traffic Generator 1 P5 P1 P2 P2 P3 P3 P4 P1 P6 P5 P6 P7 P7 P8 P8 P4 W1 W5 W2 W6 W4 W3 Example • Straight 7-to-7, 1 Gen.

  13. Algorithm • f = 1 • for (k = 1 to r, step 1) • { if(k>1) • for (j = 1 to m, step 1) • { if(j>1) f = mod*(CH(1,j-1,k)+1, TNW); • for (i =1 to NW(k), step 1) • { • CH(i,j,k)=W(f); f = mod*(f+1, TNW); • } end for i • } end for j • } end for k.

  14. Algorithm Rules • Each chain generally goes sequentiallyfrom wire i to wire i+1 unless the wire has already been fully used by other chains. Use modulo N arithmetic. • Multiple Chains/Generator: Each new VCC chain is obtained from the previous one shifting by one its wire number • Multiple Generators: Divide the wires between the generators. Each generator will start its traffic from its wires.

  15. Throughput and Latency Measurements • Performance testing requires two kinds of virtual channel connections (VCCs): foreground VCCs (traffic that is measured) and background VCCs (traffic that simply interferes with the foreground traffic). • We need in throughput measurements foreground traffic, and both foreground and background traffic in latency measurements.

  16. Throughput and Latency Measurements (Cont) • Foreground traffic in Latency measurements uses only two ports, one source and one destination. • For scalable configurations in Latency Measurements, foreground and background traffic share ports in opposite directions.

  17. 8-to-8 Straight

  18. IN OUT P2 P5 P1 P1 P3 P4 P4 P3 P5 P6 P2 P7 P8 P8 P7 P6 W1 W6 W3 W7 W2 W4 W5 8-to-8 Straight, 1 Gen. Throughput Foreground Traffic VCC chain is: P1-W1-W2-W3-W4-W5-W6-W7-P1.

  19. 8-to-2 Partial Cross Throughput Foreground

  20. IN OUT P2 P5 P1 P1 P3 P4 P4 P3 P5 P2 P6 P7 P8 P8 P7 P6 W2 W6 W1 W3 W5 W4 8-to-2 Partial Cross, 2 Gen. Throughput Foreground Traffic Gen.1 1) P1-W1-W2-W3-P1 2) P1-W2-W3-W4-P1 Gen.2 1) P2-W4-W5-W6-P2 2) P2-W5-W6-W1-P2

  21. IN OUT P2 P5 P1 P1 P3 P4 P4 P3 P5 P2 P6 P7 P8 P8 P7 P6 W4 W2 W5 W3 W1 7-to-2 Partial Cross, 2Gen. Foreground Traffic Background Traffic Gen.1 1) P1-W1-W2-W3-P22) P1-W2-W3-W4-P2 Gen.2 1) P3-W4-W5-P3 2) P3-W5-W1-P3

  22. P3 Out P3 In Modifications to Sections 3.1.7 and 3.2.7 • Existing Figure 3.3:8-to-8 straight configuration, 1 Gen. P7 In P7 Out P2 In P6 Out P2 Out P6 In ATM Monitor P1 Out P5 In P1 In P5 Out P4 Out P0 In A Out P4 In P2 Out P0 Out A In

  23. P7 P4 P2 P1 P3 P3 P4 P6 P1 P5 P6 P2 P7 P8 P8 P5 W1 W3 W5 W4 W7 W6 W2 New Figure 3.3 IN OUT G A

  24. Modifications to Sections 3.1.7 and 3.2.7 (Cont.) • Similarly, change Figure 3.4 and 3.5 • Replace “loopbacks” by “wires” • Exchange “w” and “n” for consistencyThroughout the document: n=# of ports.

  25. Summary • New Methodology: • Allows both loopback and non-loopback external connections. • Allows any number of generators.Þ Can be used for both scalable and basic configurations. • Algorithm can be implemented as a computer program.

  26. Motion • Adopt the text of 97-1089 as Appendix B of Performance Testing Baseline Text. • Adopt the appropriate modifications to Section 3.1. and 3.2.7 of the Baseline text.

More Related